The neural-morphic computing technology based on non-volatile memory electronic synapses is one of the key technologies supporting the development of artificial intelligence and has garnered significant attention in recent years. The materials and characteristics of non-volatile memory devices directly affect their performance in electronic synapse emulation and neural-morphic computing applications. Therefore, optimizing electronic synapse devices suitable for neural-morphic computing technology has become the subject of intense research, with the key challenge lying in simulating both long-term and short-term synaptic plasticity with high reliability over time.

Phase change random access memory (PCRAM) based on germanium antimony telluride (Ge2Sb2Te5, GST) is widely used in electronic synapse research due to its compatibility with integrated circuit silicon processes, mature processing, high device reliability, and moderate resistance change speed. However, current GST-based PCRAM devices typically only simulate long-term synaptic plasticity, limiting their application in the field of electronic synapses and neural-morphic computing.

Recently, Professor Ren Wei and Professor Niu Gang from the School of Electronics and Engineering of Xi'an Jiaotong University (XJTU) achieved high-reliability long-term/short-term electronic synapses by modifying traditional GST devices through ruthenium (Ru) doping.

Using the same RuGST device, this study achieved short-term/long-term synaptic plasticity, which is important for the development of PCRAM in areas such as complex cognition and higher-order learning in neural-morphic computing technology.

The achievement, titled "Long-term and short-term plasticity independently mimicked in highly reliable Ru-doped Ge2Sb2Te5 electronic synapses," was published in InfoMat . Dr Wang Qiang from XJTU is the first author of the paper, with professors Ren and Niu serving as two of the corresponding authors. XJTU's academician Jiang Zhuangde, Professor Zhao Libo, Professor Zhang Bin and doctoral student Wang Yachuan are co-authors of the paper.